Organisational information

▪ To enable audio, activate your PC speakers or connect

headphones to your PC.

▪ All participants' microphones are muted during the

webinar.

▪ If you have any questions, please use the 'Questions and

answers' window.

▪ Questions will be answered at the end of the presentation.

▪ The webinar is being recorded and will soon be made

available on our website together with the presentation
materials – you will then receive an e-mail notification.
© HBM

HBM: public 1
 Welcome to the webinar:
“Measurement Uncertainty of Force Measurements”

HBM: public 2
 Presenter

Thomas Kleckers

▪ Product Manager Industrial

Measurement (IMS) at HBM
▪ Degree in Engineering
▪ 16 years of experience in development
(strain gauges)
▪ Product manager for force sensors
▪ E-Mail: thomas.kleckers@hbm.com Thomas Kleckers
© HBM

HBM: public 3
 Agenda

- General remarks / definitions

- Accuracy classes
- Systematic errors
- How to estimate a measurement uncertainty
- Example
- Not precise enough?
© HBM

HBM: public 4
 Acccuracy – why?

• New more strict regulations and

standards require lower
measurement uncertainty

• End-of-the-line-Tests: lower
tolerance optimize the yield of
your production

=> Lower measurement uncertainty

helps to reach goals and ensure that
the measurement equipment can be
© HBM

used for years

HBM: public
 Terms and definitions

Prof. Werner Richter:

„A measurement result without an
uncertainty calculation is so much
disputable that it should not be mentioned

Kleckers
It is important to know
- what the value of my measurement uncertainty is
- how I can improve my accuracy
© HBM

HBM: public
© HBM
Terms and definitions

HBM: public
 Terms and definitions
Resolution

This measurement device has a

resolution of 1 mm

A DMP41 has 2 Mio digits in a measurement

range of +/- 2.5 mV/V
Resolution: 2,5mV/V/1Mio=0,0025 µV/V
© HBM

HBM: public
8
 Accuracy class

Accuracy class?

?
© HBM

HBM: public
 Accuracy class

Accuracy class ? Amplifier:

- Linearity
- Repeatability
- TCZero
- TC of amplification

Transducer:

· Linearity error
· Hysteresis
· Creep
· Zero point return
· Repeatability in unchanged mounting
position
· TCZero
· TCSpan
© HBM

HBM: public
 Accuracy class

Accuracy class ?
Everybody can do whatever he/she wants!
• No standard existing
• % of full scale
• Do not mix up with
Measurement uncertainty
Accuracy class according ISO376

• You can not compare sensors from different suppliers

• You can not calculate any errors or uncertainties with the
accuracy class
• BUT: Choosing a DAQ-System that fits to the sensor- this
works!
© HBM

HBM: public
11
 Systematic deviations

It is known if the difference is positive or

negative as well as the value of the
deviation
➔ have to be corrected

Example
The influence of the weight of
load introduction parts are used:
=> Tare your measurement chain
© HBM

HBM: public
1
 Influences on the measurement uncertainty,

What is the accuracy of my measurement chain?

Sorry, depends on ….
© HBM

HBM: public
13
 Influences on the measurement uncertainty,
Other measurement errors (not systematic)
It is not known if the error is positive or negative
as well as the value
➔ Measurement uncertainty

Process
Measurement chain

Measurement uncertainty

Post process or real

Calibration
time calculation

Adjustment of the measurement chain

© HBM

HBM: public
 A way to estimate the measurement uncertainty realistically

(Accuracy class) X Use of the GUM

measurement value standard

A down to earth way of calculating the

individual measurement uncertainty
© HBM

HBM: public
15
 A way to estimate the measurement uncertainty realistically

GUM =„Guide to the Expression of Uncertainty in

Measurement“

- For highest scientific demands

- complex
- High effort

„The determination of the measurement uncertainty is

not a routine job or a math's problem- a detail knowledge
about the measurement task is required“
© HBM

HBM: public
16
 A way to estimate the measurement uncertainty realistically

Methods according GUM-standard

Method A
• Get a suitable number of individual measurements
• Calculate the mean value
• The measurement uncertainty can be calculated by
calculating the standard deviation of the results

Method B
• Use of existing information on influences that have an
impact on the measurement uncertainty
• Calculating the resulting measurement uncertainty by
using the single results above

Method B is the better choice for force measurements in most

case
© HBM

HBM: public
1
 A way to estimate the measurement uncertainty

Strategy with measurement uncertainty:

- Calculation of the individual errors

- Statistical characteristic of the individual properties
- Geometrical addition
- Taking care for the range of uncertainty

We need to state: No single error is dependent on another one!

This is a more or less rough estimation

HINT: HBM Seminar “Uncertainty of measurement chains”

© HBM

HBM: public
18
 Example

Tension measurement for a component test

• Load cell U2B/5KN Capacity 5 kN

• Range of force (Sinus) between 0 and 1 kN
• Temperature range 23°C up to 45°C
• Frequency 15 Hz
• Testing duration 30 min
• Zero point setting before every test
• Adjustment according datasheet 5 kN = 2 mV/V
© HBM

HBM: public
19
 Example

Data sheet of the U2B:

• Tolerance of rated output: ± 0.2% (related to MV)

• Linearity deviation.: ± 0,1% (related to FS)
• Hysteresis 0,15 % (related to FS)
• TCSpan: ± 0,1% (realted to MV)
• TCZero: ± 0,05% (related to FS)
• Creep (30 min): ± 0,06% (related to MV)
MV = Measurement Value FS = Full scale
© HBM

HBM: public
20
 Example

Tolerance of the
rated output
(related to actual
value)

Δd C = 0.2 % of 1 kN = 2 N
© HBM

HBM: public
21
 Example

• Linearity deviation
(Related to full scale)

Δ d lin = 0,1 % of 5 kN = 5 N
© HBM

HBM: public
22
 Example

• Hysteresis
(Related to full scale) c

Fnom

Δhys = 0.15 % of 5 kN = 7,5 N

© HBM

HBM: public
 Example

• Influence of changes
in temperature on
the zero point: TCZero
(Related to full scale)

ΔTK0 = 0,05 % of 5 kN (45°C-23°C)/10K = 5,5 N

© HBM

HBM: public
 Example

• Influence of changes
in Temperature on
the Sensitivity: TCSpan
(related to actual value)

ΔTKC = 0,1 % of 1 kN (45°C-23°C)/10K = 2.2 N

© HBM

HBM: public
25
 Example

• Creep
(related to actual value)

Δcr = 0,06 % of 1 kN = 0.6N

© HBM

HBM: public
26
 Example

• Tolerance of the rated output (related to actual value)

Δd C = 0.2 % von 1 kN = 2 N
• Linearity deviation (Related to full scale)
Δd lin = 0,1 % von 5 kN = 5 N
• Hysteresis (Related to full scale)
Δhys = 0.15 % von 5 kN = 7,5 N
• TCSpan (Related to actual value)
ΔTKC = 0,1 % von 1 kN (45°C-23°C)/10K = 2.2 N
• TCZero (related to full scale)
ΔTK0 = 0,05 % von 5 kN (45°C-23°C)/10K = 5,5 N
• Creep (related to actual value)
Δcr = 0,06 % von 3 kN = 0.6N
© HBM

HBM: public
 Example

U ges  2d C + 2d lin + 2hys + 2TKC + 2TK 0 + 2cr

= (2 N) 2 + (5 N ) 2 + (7.5 N ) 2 + (2.2 N ) 2 + (5.5 N ) 2 + (0.6 N)

 10,98 N

Uncertainty: 1,1%....too big??

- Lower capacity
(lower influence of all parameters that related to full scale)
- More stable temperature conditions
(lower influence of TCZero/TCSpan)
- Calibration at HBM
(Lower linearity deviation, lower tolerance of sensitivity)
© HBM

HBM: public
 Influence of the Load cell

Repeatibility

Up to now New Improvement

0.025 0.02 25%

Kriechen
Creep

Up to now New Improvement

1.25…5 kN 0.04 0.02 50%
12.5…500kN 0.025 0.02 20% 1.25…5kN
12.5kN …500kN
© HBM

HBM: public
29
 Influence of the load cell
Linearity error

Capacity Up to now New Improvement

1.25 kN 0.03 0.02 33%
2.5 kN 0.03 0.02 33%
5 kN 0.03 0.02 33%
12.5 kN 0.04 0.03 25%
25 kN 0.04 0.03 25%
50 kN 0.04 0.035 13%
125 kN 0.04 0.035 13%
250 kN 0.04 0.035 13%
500 kN 0.06 0.05 17%
Hysteresis

Capacity Up to now New Improvement

1.25 kN 0.03 0.02 33%
2.5 kN 0.03 0.02 33%
5 kN 0.03 0.02 33%
12.5 kN 0.04 0.035 13%
25 kN 0.04 0.035 13%
50 kN 0.05 0.035 30%
125 kN 0.05 0.04 20%
250 kN 0.05 0.04 20%
© HBM

HBM: public 500 kN 0.05 0.05 0%

30
 Results not good enough?

Use the robustness of HBM´s U10M series:

200 % calibration
+ Capacity

Obergrenze in N - i.d.R. Nennkraft

100%

0 200%

100%
- Capacity
© HBM

HBM: public 31
 Influence of the Load cell

• Tolerance of the rated output (related to actual value)

Δd C = 0.02 % von 1 kN = 0,2 N
• Linearity deviation (related to full scale)
Δd lin = 0.02 % von 2.5 kN = 0,5 N
• Hysteresis (related to full scale)
Δhys = 0.02 % von 2.5 kN = 0.5 N
• TCSpan (related to actual value)
ΔTKC = 0,015 % von 1 kN (45°C-23°C)/10K = 0.33 N
• TCZero (related to full scale)
ΔTK0 = 0,015 % von 5 kN (45°C-23°C)/10K = 0.4125 N
• Creep (related to actual value)
Δcr = 0,02 % von 3 kN = 0.2N
Error: 1,03 N (=0,13%)
(2,06 N (=0,26%) in case of k = 2)
© HBM

HBM: public 32
 Additional informationen
More information can be found on our website:
www.hbm.com/force
© HBM

HBM: public 33
 HBM Academy
If you want to learn more:

On-site training: We bring the expertise to you

▪ Please contact the HBM Academy team if you are interested in
HBM Academy creating an on-site training course specifically
tailored to your company's needs.

mail: seminare@hbm.com
phone: +49 6151 8038061
© HBM

HBM: public 34
 Any questions?

Any questions?
▪ If you have any questions, please do not hesitate to contact us:
webinar@hbm.com

▪ Or email the presenter directly: thomas.kleckers@hbm.com

© HBM

HBM: public 35
www.hbm.com

HBM: public 36